TOUS DAM FAILURE DUE TO OVERTOPPING

J. Fleitz, Spain

ABSTRACT Case history category: e. The failure of Tous Dam due to overtopping during an
extreme flood event in October 1982 originated a complete review of the Spanish dam safety
regulations. Many lessons were learnt from the failure: reliable and redundant energy sources to
operate electromechanical equipment under extreme conditions, operational flood management
based on early warning systems with real-time information, emergency action plans for downstream
areas and comprehensive operation manuals including extraordinary and emergency situations have
become a standard.

TECHNICAL DETAILS

Tous Dam is located in the lower part of the Jucar River Basin about 15 km from the Mediterranean
Sea. It was designed as a multi-purpose reservoir for flood protection and river regulation for water
supply and irrigation. The first works to build a dam in Tous started in October 1958, with a project of
an 80 m high concrete dam. During construction, the geological conditions of the foundation forced to
paralyze the works in December 1964, after location of two faults in the river bed. The works were
resumed in April 1974, modifying the original design into a rock-fill dam with clay core between the
remaining concrete blocks previously built at both sides of the river bed. The modified design was the
first stage of a project with a dam crest elevation at 98.5 m to be heightened in a future second stage.
The dam structure included a large spillway in the central part provided with 3 radial gates. The total
spillway discharge capacity was 7,000 m 3/s, corresponding to the 500-year flood as required by the
current legislation. The bottom outlet had a capacity of 250 m 3/s. The reservoir capacity was 51.5 million
m3 at the normal operation level of 84.0 m and 122 million m 3 at the crest elevation of 98.5 m.

The main construction works of stage 1 finalized in March 1978 and the first impoundment began at the
same time. In November 1979 the maximum normal operation level (84.00 m) was reached and the
dam behaviour until the failure in October 1982 was satisfactory without any major incidents. Figure 1
shows the Tous Dam in operation before the failure.

Figure 36. Tous Dam before the failure

101
DESCRIPTION OF THE INCIDENT

During 20 and 21 October 1982 a particular meteorological condition consisting of a cold, high-altitude
depression surrounded by warm air with high moisture content led to extremely heavy rainfall in the
hinterland of the central Mediterranean coast of Spain. Average intensities as high as 500 mm in 24
hours were recorded in wide areas. As a result, the Júcar River basin, directly affected by the rains
upstream and downstream of Tous Dam, suffered heavy flooding all along. Particularly dramatic was
the flooding of the densely populated downstream part of the basin. Tous Dam was the last flood control
structure of the Júcar River basin, located only several kilometers upstream of residential areas. It
ultimately failed on October 20, at about 19:15 h.

That day the heavy rains quickly filled up the reservoir. The spillway gates were closed and could not
be opened as the electric network was out of order due to the weather conditions. Moreover, of the two
emergency diesel generators, one was under repair and the other one could not be started. Efforts to
raise the gates manually were fruitless.

The overtopping started at 16:50 h (see Figure 37); the maximum water level reached about 1.10 m
above the crest at 19:15 h.

Figure 37. Beginning of overtopping of Tous Dam

About 16 hours after recognizing the impossibility of operating the flood gates, the dam was overtopped
and washed out after 1 hour by erosion of a greater part of the shoulders and of the central rockfill (see
Figure 38). 300 km2 of inhabited land, including many towns and villages were severely flooded,
affecting around 200 000 people of which 10 000 had to be evacuated. There were 8 casualties and
the damages were estimated to reach $ 400 million, even if part of these damages were likely to be
caused by the floods before the arrival of the break wave.

102
 Figure 38. The remains of Tous Dam after the failure

Figure 39 summarizes the reservoir level data which were manually recorded by the responsible dam
engineer and the results of the forensic investigations carried out by the “Centro de Estudios y
Experimentación de Obras Públicas” (CEDEX), a public Spanish civil engineering research agency and
the Polytechnic Universities of Madrid and Valencia.

The main conclusions from these investigations are:

 The flood event all over the basin of the Júcar river was extraordinary: In the direct catchment
area of Tous (not affected by upstream dams) with an extension of 6,780 km 2 the average
precipitation of the event was 277 mm generating a total rainfall volume of 1,880 million m 3. In
the downstream part of the basin with an area of 3.692 km 2 the average rainfall was 226 mm
(834 million m 3). That means that even if Tous Dam had not failed, there would have been
severe flood damages.
 The estimated peak flow of the inflow hydrograph was 10,000 m 3/s equivalent to a 1 000-year
flood. In accordance with current regulations, Tous Dam had been designed for a 500-year
flood with a peak flow of 7 000 m3/s.
 The volume of the inflow hydrograph was estimated to be at least 600 million m 3, about 12
times the reservoir volume at normal operation level and 5 times at dam crest level.
 The peak inflow of approx. 10,000 m 3/s was reached after 10 pm on October 20, which is about
three hours after the beginning of the dam breach.
 The dam breach wave had a maximum peak flow of 15 000 m3/s.
 Numerical reservoir routing and physical reduced scale models showed that even with a totally
empty reservoir in the beginning of the flood event and fully open gates, the failure probably

103
 could not have been avoided. The load scenario of a 1 000-year flood was beyond the design
assumptions.

Figure 39. Inflow and outflow hydrographs and reservoir level evolution during October 20 and 21 at
Tous Dam (Source: Témez et al. and Arenillas et al.)

LESSONS LEARNT

In the context of the present bulletin on surveillance case histories, the main lesson to be learnt from
the Tous Dam failure is the utmost importance of reliable and redundant energy sources to operate all
essential devices and especially the electromechanical equipment of the outlet works, mainly under
extreme conditions.

Two engineers of the technical team in charge of the operation of the dam were condemned by the
Spanish legal authorities because of gross negligence in the longest litigation in Spanish history of
justice. The parallel engineering trial came to a different conclusion: the flood event was far beyond the
design criteria correctly adapted according to the current regulations. Even perfectly operating gates or
totally open gates would not have provided sufficient routing capacity to avoid the raise of the reservoir
beyond the dam crest and to prevent the failure.

The failure of Tous Dam had a very strong impact in the Spanish society and especially in the dam
engineering community and among responsible dam engineers. The latter found themselves
defenceless and without legal protection.

This situation originated a complete review of the dam safety regulations, enhancing design criteria
especially for embankment dams but also focusing on adequate operation procedures, regular safety
inspections and emergency action plans. Tous Dam was also the origin for implementing automatic
hydro meteorological networks throughout the Spanish River basins to collect real-time data and to
improve flood forecast and management. Further it contributed to the international debate about the

104
pros and cons of gated spillways. After the Tous Dam failure the majority of the dams in the Spanish
Mediterranean basins changed their operation manuals leaving their gates permanently open.

A new Tous Dam (see Figure 40) was built on the same site and part of the clay core material, which
had shown a relatively high resistance to water flow, was reused for constructing the new dam. The
main characteristics of the new dam compared to the one that failed (see Table 7), show the impact of
the new safety standards:

Figure 40. The new Tous Dam

Table 7. Comparison between the old and the new Tous Dam

Old dam New dam

Dam height 46.5 m 110.5 m

Crest elevation and
98.5 m / 122 million m3 162.5 m / 792 million m3
reservoir volume
130.0 m / 379 million m3
Normal operation level
84.0 m / 51.5 million m3 (90.0 m / 80 million m 3 during flood
and volume
season)
Spillway capacity 7 000 m3/s (gated) 20 000 m3/s (ungated)

SELECTED REFERENCES

Utrillas Serrano J.L., 2013, La presa de Tous, Ingeniería, seguridad y desarrollo en la Ribera del Júcar,
Confederación Hidrográfica del Júcar.

Témez J.R. & Mateos C. (CEDEX), 1993, Hidrograma de entrada a Tous. Datos para un juicio crítico,
Revista de Obras Públicas Nº 3319, Año 140, March 1993

105
Arenillas M., Martínez R. (Polytechnic University of Madrid), Cortés R., Ferri J.A. & Botella J.
(Polytechnic University of Valencia), Nuevos datos sobre la crecida del Júcar de octubre de 1982,
Revista de Obras Públicas Nº 3323, Año 140, July-August 1993

Alcrudo F. & Mulet J., 2007, Description of the Tous Dam break case study (Spain), Journal of Hydraulic
Research Vol. 45 , Iss. sup1,2007

de Wrachien D. & Mambretti S., 2009, Dam-break Problems, Solutions and Case Studies edited by D.
de Wrachien (State University of Milan, Italy) and S. Mambretti (Politecnico de Milano, Italy), WIT
PRESS 2009

106
6.5 ONE PAGE SUMMARIES OF CASE HISTORIES

107
 EL CHÓCON DAM - INTERNAL EROSION

Dam type: Embankment - 86m high

Description: Zoned earthfill with central core
Case history category: e, c, d
Main objective: Early detection of failure mechanisms
Main benefit: Correct and on time remedial action
Observations: Early detection of potential internal erosion and assessment of remedial works
A continuous increase of piezometric levels in the rock-core contact on right abutment was observed,
leading to the design of the following remedial works: a grouting program considered necessary at both
the right and left banks as well as for the entire foundation in order to reduce the potential internal
erosion of the clay core into open foundation joints.

Figure 41. Cross section of dam

Thus, the objective of the remedial works was to reduce the seepage through the open foundation joints,
caused either by gypsum dissolution or as a result of the original grouting curtain not being completely
effective.

LESSONS LEARNT

The importance of having reliable and well-located instruments at the rock contact interface.
Piezometric measurements at the rock contact interface facilitated the detection of the increase in
piezometric pressures .

Pro-active surveillance of dams during their life cycle became the standard. This problem detected by
effective monitoring lead to further investigations and remedial works also highlighting the importance
of using with an Independent Panel of Experts.

Proper instrumentation is not just essential for monitoring during the foundation remedial work but also
for the future long-term behaviour of the remedial work. The design of the remedial works included an
important group of piezometers to check the short as well as long term efficiency of the new grouting
and drainage curtains.

108
 DURLASSBODEN DAM

Dam type: Embankment – 80 m high

Description: Zoned earthfill with central clay ore
Case history category: c, d
Main objective: Surveillance of the grout curtain
Main benefit: Good understanding of dam behaviour
Observations: Importance of the long term behaviour of deep dam foundations
Durlassboden Dam is an embankment dam with a central clay core and is founded on erodible alluvial
deposits with a maximum depth of 130 m. Due to economic and technical reasons the grout curtain was
not carried out to the full depth of the alluvium, but only down to a silt layer at a depth of about 50 m.
Therefore, water can pass through the “window” below the silt layer.

Figure 42. Cross section of the dam

These adverse foundation conditions required an extensive monitoring system that consists of relief
wells at the downstream toe, measuring weirs for measuring discharge and turbidity in the area
downstream of the dam, piezometers in the area downstream of the embankment and piezometers
upstream and downstream of the grout curtain (see coloured lines in Figure 42 above)

In 2010, after more than 40 years of operation the question was posed whether an upgrade of a part of
the area downstream of the dam or even of the grout curtain is necessary. To get the right answer to
this question an intensive evaluation of the measured data was carried out. Having data over a period
of more than 40 years should provide answers on whether any positive or negative trends could be
identified. The evaluation showed clearly that there are no negative trends.

LESSONS LEARNT

The evaluation demonstrated the satisfying long-term performance of the dam and the grout curtain
over the decades. Any necessary improvements may be based on current and future readings.

109
 GMUEND DAM

Dam type: Gravity dam – 37 m high

Description: Concrete
Case history category: a
Main objective: Influence of material of instruments on results
Main benefit: Improvement of measurement results
Observations: Totally different measurement results after change of instrument material
Gmuend Dam was initially constructed in 1943-45 as a concrete arch dam. Due to stability problems at
the left abutments a gravity dam was added downstream of the arch dam in 1964. A further improvement
and slight raising of the dam was carried out in 1991. The dam is located in the Austrian Alps at an
elevation of about 1,200 m a.s.l. The annual ambient temperature variation amounts to about 40°C,
from -20° to +20° C.

Figure 44. Cross section

Figure 43. View from downstream

The left flank downstream of the dam is now stabilized by a retaining wall with 52 anchors and it is
monitored, amongst others, with two extensometers with lengths of 12 m and 31.5 m. Both of them
were equipped with ordinary steel bars and the readings showed an annual variation of about 1 mm
with the minimum of the measured distance in autumn and the maximum in spring-time.

To detect the influence of the temperature variation on the measuring system (the extensometer), the
steel bar of the 12 m long extensometer was replaced by a bar made of invar in 2006. Invar was selected
because it exhibits a coefficient of thermal expansion of about 1/10 of that of steel. This resulted in a
significant change of the measurement results.

LESSONS LEARNT

The annual variation of the invar-extensometers is now only about 0.4 mm with the maximum in autumn
and the minimum in spring-time. Obviously, the measuring instrument is affected by the ambient
temperature far more than the subject (the deformation of the rock), which should be measured.

110
 ZILLERGRÜNDL DAM

Dam type: Arch dam

Description: Double curvature concrete
Case history category: a, b, c, d
Main objective: Surveillance of dam foundation
Main benefit: Controlled uplift in the dam foundation
Observations: Good correlation of predicted and measured dam behaviour
In the valley bottom and in the abutments of the 186 m high Zillergründl Dam, the dam is provided with
an upstream apron which reaches up to a little more than half the height of the abutments. In addition
to this, a flexible movement joint was provided in the valley bottom on the upstream side close to
concrete foundation interface in the valley bottom. With appropriate drain pipes the water pressure in
the movement joint is limited to elevation 1 745 m a.s.l, and below the base gallery to elevation
1 700 m a.s.l. Figure 45 shows a cross section of the dam and Figure 46 shows the base section in
more detail together with the uplift pressure along the base joint.

Figure 46. Cross section of dam foundation with uplift

pressure distribution - condition since 1996

Figure 45. Cross section

The Zillergründl Dam, especially the base area of the dam, is comprehensively equipped with
instruments, all in all with more than 1 000 measuring devices. This enables a comparison of measured
and calculated/predicted behaviour.

LESSONS LEARNT

This case shows that measurements with suitable monitoring equipment at a suitable location can
confirm or help to adjust calculation methods. Although predicted that the flexible movement joint would
remain in contact even at a full reservoir, the measurement results on the other hand showed that the
joint opens 6 mm, as mentioned above. A nonlinear FEM analysis carried out later, showed a dam
behaviour in accordance with the results of the measurements. Observations from a seismic event in
2011 also confirmed the validity of the nonlinear model when comparing the observed and calculated
modal frequencies.

111
 COMOÉ DAM - EROSION IN THE FOUNDATION

Dam type: Embankment dam - 25m high and 1165m long, founded on lateritic ground
Description: Homogeneous earthfill
Case history category: b, c, e
Main objective: highlight the specific behaviour of tropical soils and the importance of proper
interpretation of monitoring results
Main benefit: improving the safety of the dam without emptying the reservoir
Observations: large voids developed by erosion in the dam foundation, where canaliculus where
observed during construction
A continuous increase of uncontrolled seepage flow on both flanks, collected downstream of the dam
along with subsidence development area downstream of the toe of the left bank, was attributed to
internal erosion occurring in the foundation. A deep foundation treatment was carried out from the crest
of the dam. It included sheet piling, tube à manchette grouting and compaction grouting.

Figure 47. Cross section of dam (seepage concentrated in the silty clay horizon

The objectives of the foundation treatment were (1) to get a positive cut-off in the most vulnerable zones
of the foundation and (2) to fill the voids developed in the silty clay horizon under the downstream shell
of the embankment, on both flanks. The main indicator of the effectiveness of the treatment is the
decrease of the seepage low and downstream hydrostatic pressures. The original treatment was
concentrated on the central part of the dam where the foundation behaviour was satisfactory.

LESSONS LEARNT

Lateritic foundation may be subjected to piping even under small hydraulic head. It is now established
that under some conditions, “laterite karst” may develop, mainly by erosion.

Proper interpretation with graphical representation of statistical analysis results is very useful in
highlighting some parameters particularly for non-initiated owners.

112
 CAMEROUN SONG LOULOU DAM – AAR

Dam type: Embankment with gravity spillway section – 35m high with 8x48 MW hydropower units
Description: Earthfill and concrete
Case history category: c, e
Main objective: Effective monitoring of concrete expansion due to AAR development.
Main benefit: Monitoring system had to be improved to facilitate AAR monitoring.
Observations: AAR could have a significant impact on dam safety and operation depending on the rate
and the amplitude of expansion.
At Song Loulou Dam, the mass concrete for the gravity section contained aggregate produced from
dam foundation excavation material, corresponding to a strong micaceous gneiss interlayed with
amphibolite lenses. Dmax was 50 mm and OPC content was ranging from 325 to 335 kg/m³. Unconfined
compressive strength reached an average of 22.5 MPa.

The first signs of concrete swelling were observed in 1991, almost 12 years after the first concrete
placement. Concrete expansion was attributed to both alkali-aggregate reaction (AAR) and internal
sulphate attack.

The rate of expansion was estimated to 150 μ strain/year, according to the more reliable geodetic
observations at the crest of the intake section. The main consequences of this situation are the following:

 Threat on the safety of the dam:

o Development of opened horizontal cracks in the intake structure, in direct contact with
the reservoir. The increase in the uplift thus generated can lead to the failure;
o The spillway gates may be blocked. They can no longer be operated, which can lead
to the overtopping of the embankment section in case of large flooding.
 Blockage of the turbines with loss in power production; and
 Failure of the spillway pre-stressed tendons supporting the gate thrusts.

The monitoring system installed during construction operated satisfactorily only during the first years. It
had to be improved later on to get reliable data on the dam deformation, spillway gates and powerhouse
behaviour caused by the concrete swelling.

Subsequent to upgrading of the monitoring system as well as further studies and laboratory testing, it
was concluded in 2012-2013 that the swelling due to AAR is nearing its end. This made it possible to
consider carrying out rehabilitation works to spread out over a period of approximately 10 years at the
total cost of € 110 million. It includes further monitoring system improvements.

LESSONS LEARNT

AAR could have a significant impact on dam safety and operation depending on the rate and the
amplitude of expansion.

A monitoring system, well designed, properly constructed and maintained in a good condition is of great
help in the early detection and understanding of the AAR phenomenon.

113
 DETECTION OF WATERLINE LEAKS IN A CONCRETE FACED ROCKFILL DAM

Dam type: Embankment - 27m high

Description: Concrete face rockfill
Case history category: b, c
Main objective: Detecting sources of leakage at joints between concrete slabs
Main benefit: Target remedial actions
Observations: Long term weir flow measurements
The Kootenay Canal facility is located on the Kootenay River near the city of Castlegar in British
Columbia, Canada, and was completed in 1976. The facility consists of a 5 km long canal leading to a
30 m high concrete intake dam with four penstocks. The forebay includes two concrete faced rockfill
dams roughly 27 m in height on the north and south side of the canal. The forebay slopes are lined with
concrete slabs founded on rockfill. Leakage through the embankments is believed to originate at slab
joints. Remedial work including a geomembrane liner and a filtered drain at the downstream toe was
completed in 2009, but leakage could still be measured in the weir L4 downstream of the concrete
intake dam.

It was postulated that analysis of the correlation between forebay water level and Weir L4 flow could
provide guidance on leak elevations. The analysis was completed using an energy balance orifice flow
equation: Q = CD ∙ A ∙ √2 ∙ g ∙ (H − z)

The useful insight gained from the analysis is that only high elevation defects can produce large
changes in flow for relatively moderate changes of forebay elevation, as illustrated in the Figure below.

LESSONS LEARNT

For new construction, rigorous documentation of as-built details for features such as curb walls and
conveyance pipes may be of high value for later use by surveillance personnel. For example, at
Kootenay Canal uncertainty regarding the terminus of a buried curb wall led to years of speculation
about and investigation of the embankment flow regime.

The results of persistent, long-term investigations by surveillance personnel may have unexpected
applications. The analysis described herein, which contributed to optimization of the final design of the
2014 geomembrane, would not have been possible if the sequence of events leading to the Weir L4
upgrade had not occurred.

The proportionality of orifice flow to the square of the driving head is such that defects that develop in
low head (shallow) locations can contribute significantly to leakage through concrete-faced rockfill
embankments.

114
 INSTRUMENT PERFORMANCE EVALUATION USING TRENDS ANALYSIS

Dam type: Embankment - 180m high.

Description: Zoned earthfill with central clay core
Case history category: b
Main objective: Model relationship between an instruments and its key driver variable
Main benefit: Improve future dam performance predictions
Observations: Long term weir flow measurements
Trends analysis can be easily implemented in an Excel Spreadsheet to model the relationships between
an instrument and its key driver variable. As an example, the relationship between instrument and
reservoir level readings can be expressed as P(t) = A0 + C0*(t - t0) + A1*R(t - tg), where, P(t) = the
instrument reading at time t, R(t – tg) = the reservoir level reading at time (t– tg); tg = the time lag between
the instrument and reservoir readings (i.e. travel time of flow from the reservoir to the point where the
instrument measured); C0 = the rate of change (creep) of instrument readings with time; t0 = a reference
time for creep correction; A0 and A1 are constants.

WAC Bennett Dam is a 180 m high, 2 km long earthfill dam, located on the Peace River in north eastern
British Columbia, Canada. Weir 6, which measures seepage flows from the Right Bank Terrace and
downstream of the 1996 Sinkhole, is an important instrument in regard to dam performance. The figure
below shows recorded and predicted flows of Weir 6 and reservoir level readings since the dam was
constructed.

LESSONS LEARNT

Trends analysis is a useful tool to analyse historical instrumentation data and can be used to evaluate
the performance of an instrument.

The importance of incorporating lag time and creep in the trends analysis was emphasized. The
relationship between a driver and instrument has greatly improved by incorporating the lag time and
creep. A narrow and more precise performance bound can be obtained to predict the future dam
performance.

Consideration should be given to separate other factors from the correlation equations in a trends
analysis. Other factors such as precipitation, run-off or instrument errors may skew the correlation. A
more complex model incorporating other important factors is recommended if necessary.

115
 UNDERWATER INSPECTION OF WAC BENNETT DAM

Dam type: Embankment - 180m high.

Description: Zoned earthfill with central clay core
Case history category: c, d
Main objective: Identify and monitor features in the underwater portion of the dam
Main benefit: Create a full 3-dimensional model of the upstream surface of the dam
Observations: Simultaneous multi-beam and side scan baseline surveys
In 2013, BC Hydro undertook a comprehensive remote underwater inspection of the entire underwater
surface of the WAC Bennett Dam, located on the Peace River in north-eastern British Columbia,
Canada. The dam is a zoned 183 m high earthfill embankment that retains the 360 km long Williston
Reservoir.

The objective of the survey was to establish a highly accurate baseline surface to (i) identify and
characterise previously unknown features on the dam surface and (ii) to establish performance criteria
and methodology for underwater monitoring.

The major work items included the following: a simultaneous multi-beam and side scan baseline survey,
on-site multi-beam data analysis and reporting to identify features of interest, high-resolution re-
inspection and characterisation of the features of interest, data integration into 3D GIS and CAD models.
The result was a complete, high quality and accurate 3D surface of the underwater portion of the earth
fill dam as shown in Figure 48.

Figure 48. Image of 3D dam surface produced from multibeam data

LESSONS LEARNT

The WAC Bennett Dam underwater survey and inspection was successful at producing data of sufficient
quality and accuracy to allow the identification of previously unknown features over the coverage area.
This work demonstrated that, for the most part, the industry and underwater technology is at a state at
which underwater surveys can be performed to significant water depths with enough accuracy for dam
surveillance and monitoring. This provides benefits over historical piecemeal inspections and can be
used to provide comprehensive qualitative data over a large area.

When registered into a coherent data set, the data collected using the described underwater ROV
collection methods provided an accurate representation of the dam surface. Features within the point
cloud data were easily identifiable through the use of point cloud viewing software. The use of more
specialized GIS software can improve the ability to assess a feature and compare it with historical and
other information. Higher resolution scans using the same multibeam sensor but flying the ROV at lower
altitude and at smaller line spacing together with stationary high-resolution sonar scans provided
excellent detail of features of interest allowing for accurate characterisation and future monitoring.

116
 HETEROGENEITY OF AN EMBANKMENT DAM CORE

Dam type: Embankment – 94.5 m high.

Description: Sand/gravel fill with central till core
Case history category: a, b
Main objective: Assessing a zone of higher hydraulic conductivity in the central till core
Main benefit: Decide if remedial actions were required
Observations: Using construction control data and as-built reports
The surveillance of a 94.5 m high sand and gravel fill embankment dam in northern Quebec, Canada,
by means of temperature monitoring revealed the existence of a zone of higher hydraulic conductivity
in its central till core.

The presence of a more pervious zone may indicate cracking and/or ongoing internal erosion
phenomena which could be detrimental to the safety of the dam. Its presence can also be related to
variations in till properties due to construction procedures.

Field tests were done on the compacted till for construction control. These included mainly the
determination of grain-size distribution, water content and density. The spatial continuity of the
measured fines content in the dam during construction was computed using geostatistics and used to
predict values at unsampled locations and for the entire core volume. The hydraulic conductivities were
subsequently inferred.

The representativeness of the inferred hydraulic conductivities was then assessed with temperature
monitoring data. The figure below shows the annual temperature amplitudes contour lines superposed
with the hydraulic conductivity layering in the core.

Figure 49. Inferred hydraulic conductivities and line of maximum advection

The heterogeneities and zones of higher hydraulic conductivity in the dam core were found to be mainly
due to variations of fines content and till fabric, which depends on water content and density that
occurred during construction. No internal erosion or cracking phenomena are present in the core. No
other remedial actions are needed.

LESSONS LEARNT

Construction control data and as-built reports are essential for dam safety monitoring data interpretation.
This background information provides the historical context for interpretation. The effective synthesis of
this information by means of geostatistical analyses can help the dam owner in determining the
necessity for remedial works.

The variability of soil properties must be considered in dam safety assessments. The knowledge of the
geological deposition mechanisms of materials used in dams and the effects of construction practices
on their geotechnical properties can explain field observations and measurements that could be
otherwise attributed to detrimental effects such as internal erosion or cracking.

117
 SEEPAGE DETECTION IN AN EMBANKMENT DAM FOUNDATION USING MODERN
FIELD AND NUMERICAL TOOLS

Dam type: Embankment dam - 15 m high

Description: Sand/gravel fill with central till core
Case history category: c, e
Main objective: Minimize internal erosion in an underlying sand layer
Main benefit: Design of targeted relief wells as stabilizing measure
Observations: Electromagnetic and Lidar surveys as well as geostatistical analyses
Significant seepage was observed downstream from an embankment dam founded on a clay layer
underlain by sand located in northern Québec, Canada. This structure is a 15 m high and 1 200 m long
embankment founded on marine clay underlain by a fine sand deposit of up to 70 m in thickness.

The risk of internal erosion of the sand layer through the overlying clay is increased where the factor of
safety against uplift is lower. The calculation of factors of safety against uplift requires an estimation of
clay thickness and pore pressures for the entire area. A global electromagnetic survey was realized to
detect the main sources, pathways and exits of seepage from the sand layer. An airborne LIDAR survey
was performed to determine land surface elevations. A number of boreholes and cone penetrations
tests were realized over the years to assess the foundation stratigraphy and open-tube piezometers
were installed. Clay-sand contact elevations and piezometric levels in the sand layer were estimated
using geostatistics. A factor of safety against uplift was then computed for the entire downstream area,
as shown in Figure 1. The design of stabilizing measures was specifically targeted at the more critical
areas where factors of safety was lower. A number of relief wells were installed and their drainage effect
caused a decrease of piezometric levels in the sand layer in the lake area. This decrease of pore
pressures increased the factor of safety against uplift to acceptable values and thus decreased the
likelihood of internal erosion.

Figure 50. Minimal factors of safety against uplift

LESSONS LEARNT

Regular visual inspections are of utmost importance in the detection of adverse phenomena.
Inspections, sometimes performed in difficult conditions, provide the earliest warnings regarding the
evolution of dam behaviour. A thorough knowledge of the potential failure mechanisms pertaining to a
dam is also required to evaluate the significance of observations.

A dam owner has to be aware of the latest field and numerical tools to complement and synthesize
available data. The use of electromagnetic and laser surveys as well as geostatistical analyses were
instrumental in defining the limits of an otherwise spread-out problem and to provide an optimal solution,
in terms of costs, feasibility and effectiveness, for the long-term stabilization of the foundation.

118
 ASSESSMENT OF THE PORCE II DAM

Dam type: Gravity with embankment on right flank - 122m high

Description: Roller Compacted concrete (RCC) and homogenous earthfill
Case history category: c, d, e
Main objective: Evaluate the safety of the dam using the results of the monitoring system
Main benefit: Detect and timely attend deficiencies in the dam
Observations: The evaluation of the monitoring results facilitate the evaluation of the safety of the dam
After 12 years of operation, EPM decided to carry out an assessment of Porce II Dam (see Figure 51)
to evaluate the dam stability, to verify the design assumptions and to compare the design criteria with
the current state of the art. These different evaluations but especially the analysis of the monitoring
results allowed for the comparison of the performance of the dam with the original design assumptions
and made it possible to identify any component that does not comply with current design standards.

Porce II Dam has a total 278 instruments to measure seepage, pore pressures, deformation,
temperature and acceleration, located in the body of the dam and its foundations and adjacent areas.
This case history presents some of the most important aspects obtained from the evaluation of the
monitoring results from 2001 to date.

Figure 51. General view of Porce II Dam

From the analysis of the monitoring results and its inputs into numerical stability models it was possible
to make conclusions about the safety of the dam including its stability under both static and well as
dynamic conditions. In addition, the analysis of the monitoring results allowed to detect changes in the
records to give warnings and evaluate if anomalous behaviors were presented; so that they could take
timely corrective measures. It was concluded that the behavior of the dam is adequate according to
current criteria of dam engineering and its level of safety is satisfactory.

LESSONS LEARNT

The monitoring results were useful to determine the alarm thresholds for either the initiation or the
update of the emergency action plan.

Thanks to the assessment study it was possible to determine the need to install extra piezometers on
the earthfill dam to get data for a better understanding of the behaviour of this part of the dam.

119
 PORCE DAM (COLOMBIA): METHODOLOGY FOR THE DEFINITION OF MONITORING
THRESHOLDS

Dam type: Gravity with embankment on right flank - 122m high

Description: Roller Compacted concrete (RCC) and homogenous earthfill
Case history category: c, e
Main objective: Definition of monitoring thresholds
Main benefit: Diagnose and prevent the development of possible failure modes
Observations: The definition of alert or alarm thresholds is a fundamental basis for identifying
emergency situations that may affect the safety of the dam.
Within the procedures for risk management inherent in a civil structure, such as a dam, it is essential to
constantly monitor of all the influential variables in the performance of the dam. This will warn of the
presence of any potential threat so that corrective measures can be taken in time to avoid the
occurrence of a failure mechanism. It important to note that proper interpretation of monitoring
information as well as the observations from the visual inspections are essential to properly understand
the behavior of a dam. The results are also used to diagnose and prevent the development of possible
failure modes when monitoring information exceed a preset alarm level or threshold.

This case history describes the methodology used to define monitoring thresholds and explain why the
methodology typically applied in some parts of Europe, where recorded variables are correlated with
external variables cannot be applied successfully in tropical countries and therefore it is necessary to
propose another methodology to define thresholds directly on the statistics data of the monitoring
variables.

The proposed methodology for the analysis of monitoring information, the definition of monitoring
thresholds and their possible correlation with the potential failure modes are provided in the following
sequence:

 Compilation of monitoring information;

 Debugging information;
 Statistical information processing: Correlating the monitoring variables with the external
variables and performing statistical analysis directly on the monitoring variables;
 Definition of monitoring thresholds it includes both quantitative (monitoring results from
instrumentation) as well as qualitative (visual inspection) aspects; and
 Correlation of monitoring thresholds with previously identified potential failure modes.

LESSONS LEARNT

An alternative method for the definition of thresholds has been presented, based on statistical analysis
of the monitoring information, which departs from the classic analysis of the correlation of the monitoring
variables with the external variables, basically due to the low correlation that is typical for tropical
countries like Colombia. In its absence, the method makes an explicit separation of random and
epistemic uncertainties, and proposes the use of direct statistical analyses of the data for the definition
of the thresholds.

Once the thresholds have been defined using both quantitative (monitoring results from instrumentation)
as well as qualitative (visual inspection) means, these are then correlated with the potential failure
modes.

The proper surveillance and subsequent analysis constitute the basic tool to improve the understanding
of the potential failure modes.

120
 SANTA RITA DAM (COLOMBIA): DETECTING ANOMALOUS BEHAVIOR OF PORE
PRESSURES IN EARTHFILL DAMS AND ASSESMENT STUDY

Dam type: Three embankments – 51.5 m high.

Description: Earthfill – silty core at main embankment
Case history category: e, c, d
Main objective: Early detection of failure mechanisms.
Main benefit: Evaluate the effectiveness of the rehabilitation or update works.
Observations: Reduce of liquefaction potential and assessment of update works.
During the regular visual inspections carried out at the dam, erosion on the upstream slope and wet
areas downstream of the embankment were observed. It was also observed that the body of the dam
was saturated and the seepage was flowing through the embankment. In the area downstream of the
embankment some wet areas were observed. These anomalies led to an assessment study to
determinate the safety of Santa Rita Dam.

Figure 52. Upgrades at Auxiliary II embankment of Santa Rita Dam done in 2004

The assessment study concluded that a high liquefaction potential in Auxiliary II embankment that could
generate mass movements during the occurrence of an exceptional earthquake. The solutions
proposed by the assessment were focussed on the reduction of the liquefaction potential of soil, the
control of seepage flow through the left abutment and the reduction of the effect of possible increases
in the seepage flow through this abutment during an earthquake. The update works were carry out in
2004. The main element of the upgrade to reduce the liquefaction potential, a group of gravel wells
were installed to increase the effective stresses in the most susceptible areas. To protect the left
abutment, a blanket drain/filter and a reinforcement fill were added on the downstream side, to control
fines migration and ensure efficient drainage system. In addition, geotechnical instrumentation
(vibrating wire piezometers and seepage weirs) was installed to improve monitoring of the dam and to
verify the effectiveness of the upgrades in the long term.

LESSONS LEARNT

An efficient instrumentation program should: provide early warning about a potential problem, assist in
the evaluation of a detected problem and evaluate the effectiveness of the rehabilitation or upgrades.
Despite the initial low number of instruments in the Auxiliary II embankment, the available results still
made it possible to detect excessive seepage from the reservoir into the body of the dam, to evaluate
the criticality of the problem and also to highlight the need improved monitoring. It resulted in the
installation of more instruments, which currently continues the permanent monitoring of the area.

It is important to complement the analysis of the instrumentation with the results of visual inspections,
to have certainty about the problem evaluated. The visual inspections confirm the monitroing
observations as seepage and wet areas were observed in the same places where the instruments
recorded anomalies.

121
 TONA DAM (COLOMBIA): FIELD ROCKFILL DEFORMATION MODULUS

Dam type: Embankment - 103 meters high

Description: Concrete faced rockfill (CFRD)
Case history category: a, c
Main objective: Compare the deformation modulus of large-scale laboratory tests and available
literature with observed values
Main benefit: The laboratory test generates a good approximation of the magnitude of the rockfill
deformation modulus
Observations: A proper estimation of the deformation modulus allows one to optimize the design and
lower the risks of the design.
The design of concrete face rockfill dams requires the control of cracks in concrete elements, principally
on the face and plinth and the resulting increase in seepage flows and loss of volume of fill material. In
this case history, the monitoring results from the instrumentation were used to estimate the deformation
modulus for comparison with the those estimated from the large scale tests and those determined using
methods used during design described in literature to improve the methodologies to estimate the
deformation modulus.

This paper addresses the specific case of the Tona River Dam which is a Concrete Face Rockfill Dam
(CFRD) currently under construction for the urban water supply Bucaramanga city. The dam, 103 m
high, 248 m long with a crest width of 9 m has upstream and downstream slopes of 1.4H:1.0V and
1.5H:1.0V respectively. The total rockfill volume is about 1,9 million m3. For this case history the
process used to determine the rockfill modulus during design are described in more detail.
Subsequently, comparisons with estimates from large scale tests and those obtained from the
monitoring results from instrumentation installed during the construction of the dam were made.

LESSONS LEARNT

From literature, a wide range of values for the deformation modulus were estimated. Nevertheless,
these values are useful for an initial design. It is necessary to refine these values in order to execute a
more realistic design in terms of displacement estimation.

Summing up and comparing the diverse methodologies, it can be observed how the execution of large
scale laboratory testing is recommendable. Despite its high cost, it generates a good approximation of
the magnitude of the rockfill deformation modulus, which allows one to optimize the design and lower
the risks of the system.

With the method of deformation modulus scaling proposed by Saboya (2000), similar values to those
obtained by the instrumentation that was installed during the construction were obtained.

The deformation modulus values obtained from the monitoring results and those obtained by means of
scaling of laboratory test results are bigger than the ones anticipated for the design by means of
literature references; therefore, lower deformations in the concrete face were expected. As a result of
this, there was some minor cracking on the first filling of the reservoir with consequent lower seepage
flows.

With regards to instrumentation, it is recommended to install redundant types of instruments and

amounts that allow cross correlations in order to verify when there is an abnormal behaviour, whether
if it’s the installation or reading of the equipment and records or if its presenting a behaviour that requires
special attention.

122
 EROSION PRESENTED DOWNSTREAM IN A STRUCTURAL FILL IN THE RIGHT
ABUTMENT OF THE URRÁ I HYDROELECTRIC PROJECT (COLOMBIA)

Dam type: Embankment - 73m high

Description: Earthfill with clay core
Case history category: e
Main objective: Early detection of failure mechanisms
Main benefit: Correct and on time remedial action
Observations: Early detection of landslide and assessment of remedial works.
A possible slope failure in a disposal fill identified on the right abutment of the Urrá I Hydroelectric
Project because of the saturation of materials in the disposal fill. The structural fill was not compromised
at all nor the natural slope’s integrity on the abutment at this point.

Figure 53. Layout of Urrá Dam

Thanks to the observations from the visual inspections and the proper interpretation of the monitoring
results, the risk of failure was identified and mitigated through the installation of a drainage system.

LESSONS LEARNT

For any zone to be used as a disposal site especially if it is adjoining a structural fill, careful
consideration should be given to the short and long-term impacts, the compaction methods and
placement of the material otherwise future the stabilization costs will increase.

The monitoring system should be properly designed during the design and consider amongst other
things soil conditions and parameters to be measured to determine the stability of a structure over a
given period of time.

The implementation and execution of preventive maintenance schedules and regular monitoring
interpretation are very important for any conclusions pertaining to a structure’s safety.

123
 MŠENO DAM – AGEING OF MASONRY DAM FOUNDATION ZONE

Dam type: Masonry dam - 20m high.

Case history category: a, d, e
Main objective: Ageing process of dam foundation zone, remedial measures
Main benefit: Bedrock sealing made from new grouting gallery
Observations: Long-term development of seepage regime

Masonry dam was in operation from 1909. A complications during regular maintenance of the area
near the downstream toe in the 80´s indicated changes in permeability in the dam foundation zone.
There were done several geotechnical surveys to evaluate preciously bedrock, the material in
foundation zone and to monitor seepage regime by open piezometers as well. The expert evaluations
of the dam stability were made. Operator and dam safety experts carried out large in situ operational
tests to verify the influence of hydraulic and temperature loads to the deformation of the dam body.
Different ways of remedial
measures were taken into
account. Finally, more
expensive and more risky one
was chosen – excavation of the
new grouting gallery in contact
with the dam foundation zone
and short grouting curtain from
the gallery to bedrock.
Other advantages of this
solution are:
 Detailed geotechnical
knowledge of the dam
foundation area.
 More reliable monitoring of the
foundation zone and grout
curtain sealing effectiveness.
 Permanent access to
foundation zone for additional
sealing works in the future.
Remedial measure works
started in October 1998 and
finished in December 2000.

Fig. 1. Cross section of dam

LESSONS LEARNT

Crucial decision about bedrock sealing technology could be done on the basis of 3 aspects. Very
detailed geotechnical and geophysical survey, expert evaluation of dam stability including numerical
modelling as well as extra deformation measurements provided during large-scale operational tests.
Although the reservoir was emptied to the inactive storage water level during the gallery excavation
works, precise dam behaviour monitoring was necessary to control and modify the process and
technology of ongoing works.

124
 EL KARM DAM – SPILLWAY AND DAM BODY CRACKS

Dam type: Concrete Gravity Dam – 25 m high.

Case history category: e
Main objective: Early detection of cracks mechanisms
Main benefit: Correct and on time remedial action
Observations: Early detection of potential cracks and assessment of remedial works

After seven years of construction of El- Karm dam at the end of year 1998, cracks were observed
in the dam body perpendicular to the dam axis and between the dam body (plain concrete) and the
stilling basin behind the dam (reinforced concrete). The cracks have been monitored since its
appearance It highlighted the utmost importance of dam safety surveillance as well as the significance
of proper geological- and geotechnical investigations.

(223.0)

(220.0)

(217.0)

(207.0)
0.25:1.0

0.
8:
1.
0

(203.0) G.L.

(198.0)

(195.0)
1:2

1:2

Figure (1) Cross-section of the dam

Thus, the objective of monitoring of the cracks was to stop its movement and blocked it and to prevent
the seepage through the cracks.

LESSONS LEARNT

The investigation to know the reason of the cracks produced at the stilling basin walls and floor,
finite element has been carried out for the overflow section, and concluded that high tensile stress
produced at the location of connection between the dam body and the stilling basin wall due to the
differential settlement between the dam and the basin. That explains the reasons of the cracks between
the dam body and the basin. So the problem was detected and the study and remedial works were
followed with the Independent Panel of Experts.

The foundation (bedrock) should be uniform in the longitudinal direction in order to avoid the
excessive tensile stress in longitudinal direction and to avoid the differential settlement between the
dam monoliths.

The stilling basin of the spillway should be placed separately from the dam body to avoid the
formation of cracks between them.

125